PhishBlock AI

PhishBlock AI is a smart browser extension that protects users from phishing attacks in real time. It uses AI to scan emails, websites, and messages for suspicious content, warning users.

Comment

🚀 Motivation

Every year, phishing attempts get better at tricking people and businesses. We saw that even while spam filters are available, most solutions don't explain to users why a message is risky. We wanted to make a web app using PhishBlock AI that not only finds phishing threats with AI but also explains them in simple terms to teach and empower users.

⚙️ What it does

PhishBlock AI is a web software that utilizes AI to look at emails, URLs, and messages that seem fishy. Users can copy and paste email content or upload text, and PhishBlock AI will find phishing threats right away and explain the hazards in easy-to-understand language. The site is made for regular people, students, and small groups who want to learn more about cybersecurity.

🔧 How we made it

We used the following stack to make PhishBlock AI:

  • Next.js as the full-stack web framework for the front end
  • Supabase for storing data safely and verifying users
  • OpenAI API to find phishing attempts and make explanations that people can understand
  • Tailwind CSS for making a clean and easy-to-use UI
  • Vercel for hosting and deploying

The software enables users to talk to the AI in real time and get both a detection verdict and an explanation of each hazard.

🧩 Problems we faced

  • Teaching AI to spot social engineering: It took a lot of work to make prompts that might capture small psychological tricks used in phishing emails.
  • User experience: One of the biggest challenges in UI/UX design was making a complex topic like phishing easy to understand for people who aren't tech-savvy.
  • Parsing and formatting input: To make sure that users got consistent results when they pasted in unstructured text, we needed strong preprocessing logic.

🏆 Things we're proud of

  • Getting a working phishing detection app up and running in the time span of the hackathon
  • Getting the AI to not only find risks but also communicate them effectively to people who aren't specialists
  • Creating an interface that is smooth and responsive and makes security education easy to get to
  • Creating a base that can grow into the business or education markets

📚 What we learned

  • How to make LLM prompts for cybersecurity scenarios
  • The role of UX in generating trust in security tools
  • The importance of AI-powered platforms that give explanations that people can understand
  • How to safely and smoothly combine Supabase auth and OpenAI in a web app

🔮 What’s next for PhishBlock AI

  • Gmail integration so it can scan genuine inboxes using the Gmail API
  • Training simulation mode for schools and businesses
  • Team dashboard to keep an eye on threats across several users
  • Fine-tuning phishing detection models with real-world datasets
  • A mobile-friendly version to make it easier to use across devices

Built With

Share this project:

Updates